Image based optimisation without global consistency for constant time monocular visual SLAM

Vincent Wen Han Lui; Tom William Drummond

doi:10.1109/ICRA.2015.7140011

Image based optimisation without global consistency for constant time monocular visual SLAM

Vincent Wen Han Lui, Tom William Drummond

Electrical and Computer Systems Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

10 Citations (Scopus)

Abstract

This paper presents a monocular visual SLAM system that does not require a globally consistent 3D model. Instead of generating a globally consistent 3D model and localising the camera from the 3D model, the system merely optimises relative pose parameters for pairs of keyframes that overlap on the scene, providing accurate local information at the expense of global consistency. During run-time, the camera is localised using only 2D measurements from nearby keyframes instead of using correspondences between 2D measurements and 3D features of a 3D model. Extensive experiments using both synthetic and real data sets were performed to evaluate the system s performance. Results show that our system is accurate and runs in real time at an average of 25 frames per second on a standard computer. Finally, we also show how useful applications can be easily developed on top of a framework without global consistency.

Original language	English
Title of host publication	Proceedings of 2015 IEEE International Conference on Robotics and Automation (ICRA)
Editors	Allison Okamura
Place of Publication	Washington DC USA
Publisher	IEEE, Institute of Electrical and Electronics Engineers
Pages	5799 - 5806
Number of pages	8
Volume	1
ISBN (Print)	9781479969234
DOIs	https://doi.org/10.1109/ICRA.2015.7140011
Publication status	Published - 2015
Event	IEEE International Conference on Robotics and Automation 2015 - Seattle, United States of America Duration: 26 May 2015 → 30 May 2015 http://ewh.ieee.org/soc/ras/conf/FullySponsored/ICRA/2015/icra2015.org/index.html https://ieeexplore.ieee.org/xpl/conhome/7128761/proceeding (Proceedings)

Conference

Conference	IEEE International Conference on Robotics and Automation 2015
Abbreviated title	ICRA 2015
Country/Territory	United States of America
City	Seattle
Period	26/05/15 → 30/05/15
Internet address	http://ewh.ieee.org/soc/ras/conf/FullySponsored/ICRA/2015/icra2015.org/index.html https://ieeexplore.ieee.org/xpl/conhome/7128761/proceeding (Proceedings)

Access to Document

10.1109/ICRA.2015.7140011

2741099-oaFinal published version, 2.41 MB

Cite this

@inproceedings{951cf164495f4225aa5555241f04a4c6,

title = "Image based optimisation without global consistency for constant time monocular visual SLAM",

abstract = "This paper presents a monocular visual SLAM system that does not require a globally consistent 3D model. Instead of generating a globally consistent 3D model and localising the camera from the 3D model, the system merely optimises relative pose parameters for pairs of keyframes that overlap on the scene, providing accurate local information at the expense of global consistency. During run-time, the camera is localised using only 2D measurements from nearby keyframes instead of using correspondences between 2D measurements and 3D features of a 3D model. Extensive experiments using both synthetic and real data sets were performed to evaluate the system s performance. Results show that our system is accurate and runs in real time at an average of 25 frames per second on a standard computer. Finally, we also show how useful applications can be easily developed on top of a framework without global consistency.",

author = "Lui, {Vincent Wen Han} and Drummond, {Tom William}",

year = "2015",

doi = "10.1109/ICRA.2015.7140011",

language = "English",

isbn = "9781479969234",

volume = "1",

pages = "5799 -- 5806",

editor = "Allison Okamura",

booktitle = "Proceedings of 2015 IEEE International Conference on Robotics and Automation (ICRA)",

publisher = "IEEE, Institute of Electrical and Electronics Engineers",

address = "United States of America",

note = "IEEE International Conference on Robotics and Automation 2015, ICRA 2015 ; Conference date: 26-05-2015 Through 30-05-2015",

url = "http://ewh.ieee.org/soc/ras/conf/FullySponsored/ICRA/2015/icra2015.org/index.html, https://ieeexplore.ieee.org/xpl/conhome/7128761/proceeding",

}

Lui, VWH & Drummond, TW 2015, Image based optimisation without global consistency for constant time monocular visual SLAM. in A Okamura (ed.), Proceedings of 2015 IEEE International Conference on Robotics and Automation (ICRA). vol. 1, IEEE, Institute of Electrical and Electronics Engineers, Washington DC USA, pp. 5799 - 5806, IEEE International Conference on Robotics and Automation 2015, Seattle, Washington, United States of America, 26/05/15. https://doi.org/10.1109/ICRA.2015.7140011

Image based optimisation without global consistency for constant time monocular visual SLAM. / Lui, Vincent Wen Han; Drummond, Tom William.
Proceedings of 2015 IEEE International Conference on Robotics and Automation (ICRA). ed. / Allison Okamura. Vol. 1 Washington DC USA: IEEE, Institute of Electrical and Electronics Engineers, 2015. p. 5799 - 5806.

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

TY - GEN

T1 - Image based optimisation without global consistency for constant time monocular visual SLAM

AU - Lui, Vincent Wen Han

AU - Drummond, Tom William

PY - 2015

Y1 - 2015

N2 - This paper presents a monocular visual SLAM system that does not require a globally consistent 3D model. Instead of generating a globally consistent 3D model and localising the camera from the 3D model, the system merely optimises relative pose parameters for pairs of keyframes that overlap on the scene, providing accurate local information at the expense of global consistency. During run-time, the camera is localised using only 2D measurements from nearby keyframes instead of using correspondences between 2D measurements and 3D features of a 3D model. Extensive experiments using both synthetic and real data sets were performed to evaluate the system s performance. Results show that our system is accurate and runs in real time at an average of 25 frames per second on a standard computer. Finally, we also show how useful applications can be easily developed on top of a framework without global consistency.

AB - This paper presents a monocular visual SLAM system that does not require a globally consistent 3D model. Instead of generating a globally consistent 3D model and localising the camera from the 3D model, the system merely optimises relative pose parameters for pairs of keyframes that overlap on the scene, providing accurate local information at the expense of global consistency. During run-time, the camera is localised using only 2D measurements from nearby keyframes instead of using correspondences between 2D measurements and 3D features of a 3D model. Extensive experiments using both synthetic and real data sets were performed to evaluate the system s performance. Results show that our system is accurate and runs in real time at an average of 25 frames per second on a standard computer. Finally, we also show how useful applications can be easily developed on top of a framework without global consistency.

U2 - 10.1109/ICRA.2015.7140011

DO - 10.1109/ICRA.2015.7140011

M3 - Conference Paper

SN - 9781479969234

VL - 1

SP - 5799

EP - 5806

BT - Proceedings of 2015 IEEE International Conference on Robotics and Automation (ICRA)

A2 - Okamura, Allison

PB - IEEE, Institute of Electrical and Electronics Engineers

CY - Washington DC USA

T2 - IEEE International Conference on Robotics and Automation 2015

Y2 - 26 May 2015 through 30 May 2015

ER -